Efficacy and safety of neoadjuvant chemotherapy …...after treatment have a high risk of recurrence [5]. Although there is a significant association between pCR and survival outcomes,

www.aging-us.com 6286 AGING

INTRODUCTION Triple-negative breast cancer (TNBC) accounts for ~12%–17% of all breast cancers [1]. Because endocrine and anti-HER2 therapies are not suitable for TNBC patients, adjuvant chemotherapy is generally the only line of systemic treatment [1, 2]. Compared to other subtypes of breast cancer (e.g., hormone receptor or HER2 positive), TNBC patients suffer worse clinical outcomes [3]. In patients with early stage TNBC, neoadjuvant

chemotherapy has become a standard approach and is more likely to achieve pathological complete response (pCR) than non-TNBC patients [4]. Event-free survival and overall survival is dramatically improved in TNBC patients who achieve a pCR after neoadjuvant chemo-therapy, whereas patients with residual invasive disease after treatment have a high risk of recurrence [5]. Although there is a significant association between pCR and survival outcomes, the overall prognosis of TNBC

www.aging-us.com AGING 2019, Vol. 11, No. 16

Research Paper

Efficacy and safety of neoadjuvant chemotherapy regimens for triple-negative breast cancer: a network meta-analysis Yunhai Li1,2, Dejuan Yang1, Ping Chen3, Xuedong Yin1, Jiazheng Sun1,2, Hongzhong Li1,2, Guosheng Ren1,2

1Department of Endocrine and Breast Surgery, The First Affiliated Hospital of Chongqing Medical University, Chongqing, China 2Chongqing Key Laboratory of Molecular Oncology and Epigenetics, The First Affiliated Hospital of Chongqing Medical University, Chongqing, China 3Department of Rheumatology, Daping Hospital, The Third Affiliated Hospital of Third Military Medical University, Chongqing, China Correspondence to: Hongzhong Li, Guosheng Ren; email: [email protected], [email protected] Keywords: network meta-analysis, triple-negative breast cancer, neoadjuvant chemotherapy, pathological complete response Received: May 13, 2019 Accepted: August 10, 2019 Published: August 24, 2019 Copyright: Li et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT Different neoadjuvant chemotherapies are available for triple-negative breast cancer (TNBC). Here, we performed a network meta-analysis to evaluate the pathological complete response (pCR) benefit and safety of treatment regimens. Pairwise and Bayesian network meta-analyses were performed to compare direct and indirect evidence, respectively. Twenty-three studies involving 12 regimens namely standard chemotherapeutic agents, bevacizumab (B)-, platinum salts (P)-, B plus P (BP)-, poly(ADP-ribose) polymerase inhibitors (Pi)-, P plus Pi (PPi)-, capecitabine (Ca)-, gemcitabine (Ge)-, zoledronic acid (Za)-, everolimus (E)-, P plus E (PE)-, and gefitinib (G)-containing regimens. The results showed that P-, B-, PPi-, and Za-containing regimens achieved higher pCR than standard chemotherapeutic agents. BP-containing regimens had a better pCR than B-containing regimens. In indirect comparisons, Za-, BP-, P-, and B-containing regimens were the top four strategies with the highest probability for pCR. Benefit-risk analysis showed that B-containing regimens had the highest acceptability of being the best treatment for better pCR achievement with fewer SAEs. The addition of P, B, BP, PPi, and Za to standard chemotherapeutic agents enhanced the pCR, but a balance between efficacy and safety should be carefully considered. B-containing regimens might be the best choice for neoadjuvant chemotherapy due to its better efficacy and tolerability.

mailto:[email protected]

mailto:[email protected]


remains unsatisfactory, and only ~30% of TNBC patients achieve pCR following treatment with standard anthracycline-, cyclophosphamide-, taxane-, and/or fluorouracil-based neoadjuvant chemotherapy [6]. Currently, different strategies are applied to increase TNBC pCR rates. For instance, adding platinum salts, which induce double-stranded DNA breaks and sub-sequent cell death, to standard neoadjuvant chemo-therapy has been explored by several randomized control trials (RCTs), and has shown a promising pCR benefit in TNBC patients, especially in those with BRCA mutations [7–9]. However, severe toxicities resulting in hemato-logical, gastrointestinal, and nervous system disorders have frequently been observed in patients treated with platinum-containing regimens [9]. Bevacizumab, a monoclonal antibody targeting vascular endothelial growth factor A, has been demonstrated to improve pCR rates when added to neoadjuvant chemotherapy in TNBC [10]. In I-SPY 2, a multicenter phase 2 trial, the addition of veliparib and carboplatin to standard neoadjuvant chemotherapy significantly increased the pCR proportion in TNBC patients compared with controls [11]. However, the NeoPARP study found no difference in pCR between an iniparib plus paclitaxel regimen vs. paclitaxel alone in TNBC [12]. Despite these results, the optimal regimen for TNBC treatment remains controversial due to the limitations of RCTs and conventional meta-analyses in comparing and integrating the efficacy of all available regimens. Network meta-analysis, also known as multiple-treatments or mixed-treatment analysis, not only allows the integration of the evidence without head-to-head comparison, but also can determine the superiority of different interventions by ranking probability and acceptability [13]. Therefore, we conducted a Bayesian network meta-analysis to estimate the efficacy and safety of the currently available neoadjuvant chemo-therapies in TNBC. Here, we provide a systematic summary of different regimens that may aid in treatment decisions and future studies. RESULTS Overview of literature search and study characteristics A total of 2,138 relevant records were identified in the electronic databases. Of these, 82 potentially eligible studies were reviewed with full text. Fifty-nine studies were excluded because 29 were out of scope, eight were duplicated data, 16 were not suitable for analysis, four were retrospective studies, and two were single-armed studies. However, although the study by Enriquez et al. [14] is a single-armed trial, it was controlled using historical patient data; this study was therefore included

in our analysis. All together, 23 studies [7–12, 15–30] comprising 4,099 patients were included. Twenty-one studies were reported as full text publications and two [14, 25] as conference abstracts. The PRISMA flow chart of study selection is shown in Figure 1. All eligible studies were published between 2010 and 2018, and patients were enrolled between 2003 and 2017. Except for the historical controlled trial (HCT) [14], all studies were randomized controlled trials (RCTs). There were six phase 3 and 17 phase 2 trials. The pCR was defined as ypT0/is ypN0 in 20 studies, whereas in three trials [15, 17, 24] pCR was defined as ypT0/is. The main characteristics of the included studies are summarized in Table 1. Twelve treatment regimens were assessed: standard chemotherapeutic agents, bevacizumab (B)-, platinum salts (P)-, B plus P (BP)-, poly(ADP-ribose) polymerase inhibitors (Pi)-, P plus Pi (PPi)-, capecitabine (Ca)-, gemcitabine (Ge)-, zoledronic acid (Za)-, everolimus (E)-, P plus E (PE)-, and gefitinib (G)-containing regimens (Figure 2). As for the four-armed GeparTrio trial [16], we only included the patients in the two arms without early response, as the other two arms compared the efficacy between four and six cycles of standard chemotherapeutic agents. Moreover, vinorelbine in the Ge-containing arm was ignored in the GeparTrio trial. In the SOLTI NeoPARP [12] trial, we combined the two arms that used different dosages of iniparib into the Pi-containing regimen for the final analysis. The treatment details of the included trials are shown in Supplementary Table 1. The quality of evidence was evaluated using the Cochrane risk of bias tool [46]. There was low risk of bias for the majority of categories. However, caution should be taken for the HCT [14] due to its high risk of selection bias. The results of quality assessment are shown in Supplementary Table 2. Pairwise meta-analysis of primary outcome There were eight studies that directly compared standard chemotherapeutic agents with P-containing regimens. Because a modest heterogeneity (p=0.09, I2=43%) was detected, a pairwise comparison was performed with a random-effects model. We found that P-containing regimens were associated with a sig-nificant pCR benefit compared to standard chemo-therapeutic agents (OR=2.41, 95% CI: 1.65–3.51, p<0.00001; Figure 3A). B-containing regimens had a better pCR than standard chemotherapeutic agents (OR=1.60, 95% CI: 1.26–2.02, p<0.0001) as analyzed by fixed-effect model (p=0.25, I2=28%; Figure 3B). BP-containing regimens were more effective than B-containing regimens (OR=1.86, 95% CI: 1.32–2.63, p=0.0004; Figure 3C). Two studies compared PPi-


containing regimens with standard chemotherapeutic agents (OR=2.60, 95% CI: 1.78–3.81, p<0.00001), which indicated that PPi-containing regimens were superior in pCR achieving (Figure 3D); however, caution should be taken because the weights of the two studies were 90.6% and 9.4%, respectively. There was no significant difference between standard chemo-therapeutic agents and Ca-containing regimens (OR=0.94, 95% CI: 0.21–4.18, p=0.94; Figure 3E). Za-containing regimens showed a significant pCR benefit (OR=3.72, 95% CI: 1.07–12.93, p=0.04) compared to standard chemotherapeutic agents (Figure 3F). Sensitivity analyses were conducted to detect the influence of individual studies on the comparisons of standard chemotherapeutic agents with P- and B-containing regimens by omitting one study at a time. Omission of any single study did not materially alter

the pooled effects (data not shown). The shapes of funnel plots of the two comparisons were close to symmetric, and no significant publication bias was identified by Begg's and Egger's tests (Supplementary Figure 1). Bayesian network meta-analysis of primary outcome All of the 23 trials were included in the network meta-analysis for pCR. Node-splitting analysis was performed in order to evaluate the inconsistency, and no statistical difference was identified between direct and indirect evidence (data not shown). Therefore, the Bayesian network meta-analysis was conducted with a consistency model (Figure 4A). The results showed that the pCR incidence achieved by standard chemo-therapeutic agents was significantly lower than B- (OR=0.58, 95% CI: 0.39–0.80), P- (OR=0.41, 95%

Figure 1. A PRISMA flow chart of the literature search and study selection in this meta-analysis.


Table 1. Characteristics of eligible studies.

Study First author country Type of trail Trial

phase Masking Recruit

ment period

No. of center Arms TNBC

definition Clinical

stage

No. of patients analyzed

Trial name/

registry number

Aft 2010 America Prospective RCT II Open-

label 2003–2006 Single 2

ER/PR=0%, HER2=

–; 1+; 2+/Hish–

II-III 40 NCT00242203

Houber 2010 Switzerland Prospective

RCT III Open-label

2002–2005 Multiple 4

ER/PR<10%, HER2=

–; 1+; 2+/Hish–

II-III 89 GeparTrio/NCT00544

765

Bernsdorf 2011 Sweden Prospective

RCT II Double-blind

2004–2007 Multiple 2 NA II-III 82 NCT

00239343

Alba 2012 Spain Prospective RCT II Open-

label 2007–2010 Multiple 2

ER/PR≤1%, HER2=

–; 1+; 2+/Hish-

II-III 93

GEICAM/2006-

03/NCT00432172

Gerber 2013 Germany Prospective

RCT III Open-label


ER/PR<10%, HER2=

–; 1+; 2+/Hish–

II-III 663

GeparQuinto/GBG

44/NCT00567554

Ando 2014 Japan Prospective

RCT II Open-label


ER/PR<10%, HER2=

–; 1+; 2+/Hish–

II-III 75 NA

Earl 2014 UK Prospective RCT III Open-


ER/PR-NA;

HER2= –; 1+;

2+/Hish–

II-III 157

Neo-tAnGo/

NCT00070278

Gonzalez-Angulo 2014

America Prospective RCT II Open-

label NA Single 2

ER/PR≤5%; HER2=

–; 1+; 2+/Hish-

II-III 50 NCT00499603

Steger 2014 Austria Prospective

RCT III Open-label


ER/PR<10%, HER2=

–; 1+; 2+/Hish–

Non-IV 127 ABCSG-

24/NCT00309556

von Minckwitz 2014

Germany Prospective RCT II Open-


ER/PR<1%, HER2=

–; 1+; 2+/Hish–

II-III 315

GeparSixto-GBG

66/NCT01426880

Earl 2015 UK Prospective RCT III Open-


ER/PR score=0–

2/8; HER2= –;

1+; 2+/Hish–

II-III 241 ARTemis/NCT01093

235

Hasegawa 2015 Japan Prospective

RCT II Open-label


ER/PR-NA;

HER2= –; 1+;

2+/Hish–

II-III 34 JONIE

Llombart-Cussac 2015

Spain Prospective RCT II Open-


ER/PR<10%, HER2=

–; 1+; 2+/Hish–

II-III 140

SOLTI NeoPARP/ NCT01204

125

Martinez 2015 Mexico Prospective

RCT II Open-label NA NA 2 NA NA 61 NA

Sikov 2015 America Prospective

RCT II Open-label


ER/PR≤10%, HER2=

–; 1+; 2+/Hish–

II-III 433 CALGB 40603


Figure 2. Network diagram of eligible comparisons included in the network meta-analysis for pathological complete response (pCR). The node size is proportional to the total number of patients in the regimen. The width of each line is proportional to the number of studies comparing the two regimens linked by the line.

Nahleh 2016 America Prospective

RCT II Open-label 2010–2012 Multiple 2

ER/PR<1%, HER2=

–; 1+; 2+/Hish–

II-III 67

SWOG S0800/

NCT00856492

Zhang 2016 China Prospective

RCT II Open-label 2006–2012 NA 2

ER/PR<10%, HER2=

–; 1+; 2+/Hish–

II-III 87 NCT01276769

Rugo 2016 America Prospective RCT II Open-label 2010–

2012 Multiple 2 Special definition II-III 60

I-SPY 2/NCT01042

379

Enriquez 2017 Peru Prospective

HCT II Open-label 2013–2014 Single 2 NA II-III 61 NA?

Gluz 2017 Germany Prospective RCT II Open-label 2013–

2015 Multiple 2

ER/PR<1%, HER2=

–; 1+; 2+/Hish–

Non-IV 324

WSG-ADAPT

TN/NCT01815242

Jovanović 2017 America Prospective

RCT II Double-blind


ER/PR<10%, HER2=

–; 1+; 2+/Hish–

II-III 145 NCT00242203

Loibl 2018 Germany Prospective RCT III Double-

blind 2014–2016 Multiple 3

ER/PR<1%, HER2=

–; 1+; 2+/Hish–

II-III 634 BrighTNess/NCT020322

77

Wu 2018 China Prospective RCT II Open-label 2014–

2017 Single 2

ER/PR<10%, HER2=

–; 1+; 2+/Hish–

I-III 121 ChiCTR-

TRC-14005019

RCT, randomized controlled trail; HCT, historical controlled trial; NA, not available.


Figure 3. Forest plots of pair-wise meta-analyses for pathological complete response (pCR). (A) Standard chemotherapeutic agents vs. P-containing regimens. (B) Standard chemotherapeutic agents vs. B-containing regimens. (C) B-containing regimens vs. BP-containing regimens. (D) Standard chemotherapeutic agents vs. PPi-containing regimens. (E) Standard chemotherapeutic agents vs. Ca-containing regimens. (F) Standard chemotherapeutic agents vs. Za-containing regimens.


CI: 0.30–0.55), BP- (OR=0.32, 95% CI: 0.19–0.53), PPi- (OR=0.43, 95% CI: 0.24–0.70), and Za-containing regimens (OR=0.26, 95% CI: 0.06–0.92). B-containing regimens had a poor pCR than BP-containing regimens (OR=0.55, 95% CI: 0.34–0.91). P-containing regimens were significantly related to pCR benefit compared with Pi- (OR=3.16, 95% CI: 1.05–8.99) and Ge-containing regimens (OR=2.15, 95% CI: 1.26–3.93). BP-containing regimens showed a significantly higher rate pCR than Pi- (OR=3.98, 95% CI: 1.30–12.29), Ca- (OR=2.61, 95% CI: 1.02–7.16),

and Ge-containing regimens (OR=2.75, 95% CI: 1.33–6.09). Moreover, PPi-containing regimens showed a significant pCR advantage over Ge-containing regimens (OR=2.09, 95% CI: 1.03–4.69). Due to the pCR definition bias of three studies [15, 17, 24] and selection bias of the HCT [14], we further performed a subgroup analysis that excluded the regimens of the four studies. However, no significant change was observed for the positive results identified above (Supplementary Figure 2A), which demonstrated the robustness of these results.

Figure 4. Bayesian network meta-analysis for pathological complete response (pCR). (A) The league table of comparisons. Data are presented as odds radio (OR) and 95% confidence intervals (CI). An OR>1 favors the column-defining treatment, and an OR<1 favors the row-defining treatment. (B) Heatmap of the rank probability of the twelve regimens for pCR. Rank 1 represents the best treatment and rank 12 represents the worst. Rank probabilities sum to one, both within a rank over treatments and within a treatment over ranks.


The rank probability of regimens was also analyzed. As shown in Figure 4B, the three best treatments were Za-containing regimens (53%), BP-containing regimens (38%), and P-containing regimens (29%) ranked first to third in pCR achieving, while Pi-containing regimens (45%), standard chemotherapeutic agents (29%), and Ge-containing regimens (23%) proved to be the three worst treatments, occupying rank 12 to rank 10. In the subgroup analysis of regimens without Za-containing regimens, BP-containing regimens (64%) were the best treatment for pCR followed by P-containing regimens (41%) and PPi-containing regimens (30%). Con-sistently, Pi-containing regimens (47%), standard chemotherapeutic agents (27%), and Ge-containing regimens (20%) remained the three worst treatments (Supplementary Figure 2B). Meta-analysis for grade 3–4 hematological adverse events There were 11 trails with eight regimens that reported anemia, 16 trials with 10 regimens that reported neutropenia, and 13 trials with seven regimens that reported thrombocytopenia (Supplementary Table 3 and Supplementary Figure 3). We found that P-containing regimens were associated with higher risk of anemia (OR=16.49, 95% CI: 7.52–36.14, p<0.00001), neutropenia (OR=3.30, 95% CI: 1.35–8.08, p=0.009), and thrombocytopenia (OR=12.93, 95% CI: 5.91–28.29, p<0.00001) compared with standard chemotherapeutic agents (Supplementary Figure 4A, 4B, and 4D). Consistently, subgroup analyses based on TNBC patients also revealed that P-containing regimens resulted in a higher incidence of anemia (OR=15.54, 95% CI: 6.64–36.34, p<0.00001), neutropenia (OR=3.36, 95% CI: 1.04–10.83, p=0.04) and thrombocytopenia (OR=13.84, 95% CI: 6.14–31.20, p<0.00001), compared to standard chemotherapeutic agents (Supplementary Figure 5A–5C). B-containing regimens were associated with a higher risk of neutropenia (OR=1.20, 95% CI: 1.01–1.43, p=0.04) than standard chemotherapeutic agents (Supplementary Figure 4C). No significant bias was identified by funnel plots or Begg's and Egger's tests (Supplementary Figure 6). Indirect comparisons were then performed using a consistency model, as node-splitting analyses revealed no significant difference between the direct and indirect evidence for anemia, neutropenia, and thrombo-cytopenia (Figure 5). With respect to anemia, standard chemotherapeutic agents (OR=1.64E+22, 95% CI: 24.21–3.11E+67), B- (OR=6.84E+22, 95% CI: 75.89–1.49E+68), P- (OR=9.57E+23, 95% CI: 1099.95–1.75E+69), BP- (OR=1.87E+24, 95% CI: 1830.13–2.18E+69), PPi- (OR=4.35E+24, 95% CI: 4271.05–1.03E+70), PE- (OR=0.00, 95% CI: 0.00–0.00), and E-

containing regimens (OR=0.00, 95% CI: 0.00–0.01) were associated with higher incidences than Ge-containing regimens. P- (OR=0.03, 95% CI: 0.00–0.14), BP- (OR=0.01, 95% CI: 0.00–0.29), PPi- (OR=0.01, 95% CI: 0.00–0.10), and PE-containing regimens (OR=0.01, 95% CI: 0.00–0.62) induced more anemia events than standard chemotherapeutic agents. BP-containing regimens (OR=0.04, 95% CI: 0.00–0.97) induced more anemia events than B-containing regimens. Neutropenia was significantly lower in standard chemotherapeutic agents than in P- (OR=0.29, 95% CI: 0.11–0.76) and PPi-containing regimens (OR=0.04, 95% CI: 0.01–0.21). B- (OR=0.05, 95% CI: 0.00–0.38) and Ge-containing regimens (OR=12.73, 95% CI: 1.21–196.14) were associated with lower incidences of neutropenia compared to PPi-containing regimens. With respect to thrombocytopenia, P- (OR=0.03, 95% CI: 0.00–0.30) and PPi-containing regimens (OR=0.01, 95% CI: 0.00–0.24) were associated with a higher risk of incidence than standard chemotherapeutic agents. PPi-containing regimens also had a higher risk of thrombocytopenia compared to B-containing regimens (OR=0.01, 95% CI: 0.00–0.74). The rank probability for each treatment inducing SAEs showed that Ge-containing regimens (99%) were the best treatment, resulting in the lowest incidence of anemia adverse events, whereas PE-containing regimens (45%) were found to be the worst treatment (Supplementary Figure 7A). PPi-containing regimens ranked first for the highest prevalence of neutropenia (60%) and thrombocytopenia (64%) events (Sup-plementary Figure 7B and 7C). Pi- (26%) and E-containing regimens (52%) had the lowest probability of leading to neutropenia and thrombocytopenia, respectively. Stochastic multi-criteria acceptability analysis (SMAA) Benefit-risk analyses were performed by SMAA with missing preferences. Standard chemotherapeutic agents were set as the baseline. The analyses between pCR and anemia showed that BP-containing regimens had the highest acceptability (26.46%), ranking first (Figure 6A) with a CF=0.34 (Supplementary Figure 8A). Za-containing regimens (49.72%, Figure 6B) were the best treatment when considering pCR and neutropenia with a CF=0.62 (Supplementary Figure 8B). BP-containing regimens ranked first in acceptability (34.18%, CF=0.42), resulting in a higher pCR and fewer throm-bocytopenia events (Figure 6C and Supplementary Figure 8C). When considering pCR with the three SAEs, B-containing regimens were the best choice with highest rank acceptability (34.01%, Figure 6D) and a CF=0.43 (Supplementary Figure 8D).


DISCUSSION A growing number of clinical trials are being performed in order to improve the effectiveness of neoadjuvant chemotherapies in TNBC by adding different drugs to the standard chemotherapeutic agents. However, results are

controversial and remain isolated in the absence of systematic integration. Therefore, a comprehensive study was warranted to give a summary of the results from these publications. To the best of our knowledge, this is the first network meta-analysis to investigate the pCR efficacy and safety of neoadjuvant chemotherapy regimens in TNBC.

Figure 5. Bayesian network meta-analysis for grade 3–4 hematological adverse events. (A) The league table for comparisons of anemia. (B) The league table for comparisons of neutropenia. (C) The league table for comparisons of thrombocytopenia. Data are presented as odds radio (OR) and 95% confidence intervals (CI). An OR>1 favors the row-defining treatment, and OR<1 favors the column-defining treatment.


In the present study, we enrolled 23 clinical trials with 4,099 TNBC individuals assigned to 12 neoadjuvant chemotherapy regimens, aiming to identify which treatment was optimal in achieving higher pCR rates and resulting in fewer SAEs. The results of pairwise meta-analyses showed that the most highly studied P-containing regimens were significantly associated with better pCR rates, but worse SAEs, compared with standard chemotherapeutic agents. Consistently, two previous meta-analyses also revealed that platinum-based neoadjuvant chemotherapies clearly increased pCR rates compared with platinum-free neoadjuvant chemotherapies [31, 32]. Although no survival benefit was observed for platinum-based neoadjuvant chemo-therapy as pooled by two RCTs [31], many more studies with long-term follow-up are required to clarify the potential association between survival outcomes and platinum salts. TNBC was demonstrated to be more sensitive to platinum salts than non-TNBC [32], with the probable reason being that TNBC is more commonly related to BRCA mutations or homologous recombination DNA repair deficiencies [33, 34]. PARP

inhibitors can block DNA repair pathways, which are crucial for tumor cell survival in patients with BRCA mutations or homologous recombination DNA repair deficiencies [34]. Therefore, it is reasonable to speculate that PARP inhibitors might enhance the anti-tumor activity of cytotoxic agents resulting in DNA damage, such as platinum salts. However, in this study, although PPi-containing regimens significantly increased pCR rates compared to standard chemotherapeutic agents, there was no difference in efficacy between P- and PPi-containing regimens, indicating that PARP inhibitors did not enhance the effects of platinum salts. This result is consistent with the findings of BrighTNess trial [9]. Moreover, a benefit-risk analysis showed that PPi-containing regimens might be the worst treatment choice when considering pCR and SAEs. In addition, we found that Pi-containing regimens without platinum salts were not superior to any other regimen. Thus, our results do not support further investigation into the use of PARP inhibitors added to standard chemotherapeutic agents or in combination with platinum salts at the present dosage in TNBC patients.

Figure 6. Stochastic multi-criteria acceptability analysis for benefit-risk. (A) Rank probability of regimens based on synthesizing pCR and anemia. (B) Rank probability of regimens based on synthesizing pCR and neutropenia. (C) Rank probability of regimens based on synthesizing pCR and thrombocytopenia. (D) Rank probability of regimens based on synthesizing pCR and the three serious adverse events. Rank 1 represents the best treatment and rank N represents the worst. The proportion corresponds to the probability of each regimen to be at a specific rank.


Bevacizumab is another frequently studied agent in neoadjuvant chemotherapy for TNBC. It has shown clinical efficacy in prolonging progression-free-survival, but not overall survival, in metastatic TNBC [35]. In our work, we found that B-containing regimens were significantly associated with a higher pCR rate than standard chemotherapeutic agents, while only a modest correlation between B-containing regimens and neutropenia prevalence was detected. However, bevacizumab may lead to other adverse events in the circulatory, nervous or urinary systems [26, 35]. Consistent with our study, a recent network meta-analysis reported that bevacizumab plus chemotherapy significantly improved pCR of TNBC patients when compared with chemotherapy plus placebo [36]. Moreover, bevacizumab plus chemotherapy was demonstrated to be significantly associated with longer progression-free survival than chemotherapy alone in advanced/metastatic TNBC [37]. Although no significant different was detected between B- and P-containing regimens in pCR rates, the combination of bevacizumab and platinum salts (BP-containing regimens) was able to increase the efficacy of both B- and P-containing regimens, when compared with standard chemotherapeutic agents. Importantly, our benefit-risk analysis revealed that B-containing regimens might be the best treatment to achieve a relatively high pCR rate with fewer hematological SAEs. To our surprise, zoledronic acid, a nitrogen-containing bisphosphonate that induces osteoclast apoptosis and inhibits bone resorption [38], showed a significant pCR benefit when added to standard chemotherapeutic agents. No statistical difference in SAEs incidence was observed between Za-containing regimens and standard chemotherapeutic agents. Moreover, Za-containing regimens had the highest probability of being the best treatment for achieving pCR. However, caution should be taken due to the limited number of subjects in the two original RCTs investigating Za-containing regimens, with only 40 and 34 patients, respectively [15, 24]. Although the effects of zoledronic acid on pCR rate were not striking in the two RCTs, the pooled effect was statistically significant, mainly due to the larger sample size in our work. Accordingly, we hold the opinion that further RCTs with larger sample sizes are needed to confirm any potential pCR benefit of zoledronic acid, and to strengthen our results. In addition to its anti-metastatic properties within bone tissue, the anti-tumor activities of zoledronic acid may be explained by several mechanisms, including inducing tumor cell apoptosis, enhancing the cytotoxic effects of chemotherapeutic agents, suppressing neo-angiogenesis, and involving immunomodulation [39–42]. Moreover, disease-free survival benefits of

zoledronic acid have also been detected in premenopausal endocrine-positive patients [43, 44]. However, the survival benefit of zoledronic acid in TNBC remains unclear. There are several limitations of the present study. First, the sample sizes of several trials were relatively small. Specifically, the three RCTs involving Za-containing regimens and the study of E-containing regimens only included approximately 50 TNBC patients, which might weaken the effects or lead to false positive results. In addition, there were only one study for E-, PE, and Ge-containing regimens and the number of SAEs very relatively small in these studies, which might result in the wide 95% CIs; therefore, the results of SEAs regarding these regimens should be interpreted with caution. Second, there are several potential heterogeneities in the included studies, such as pCR definition, study design, and the specific components of standard chemo-therapeutic agents. However, subgroup and sensitivity analyses based on pCR definition and study design demonstrated that the heterogeneities were unlikely to refute the overall results. Moreover, we integrated taxanes, anthracyclines, cyclophosphamide, and fluo-rouracil as comparative standard chemo-therapeutic agents, as most control arms of the included studies were based on these four kinds of drugs, but differences still existed between studies. Therefore, consideration of these heterogeneities should be taken into account in the interpretation of our findings. Third, the kinds of SAEs reported in each study were different, and SAE data were not available for several trials. Thus, it was difficult to conduct a comprehensive analysis for all SAEs and regimens. Although there was no evidence showing a significant correlation between SAEs and the molecular subtype of breast cancer, there may be bias in the pooled effects of SAEs compared to the real results. Fourth, the correlation between neoadjuvant chemo-therapy and survival outcomes was not evaluated due to the lack of survival data in most studies. However, we noticed that some RCTs are still ongoing and a secondary analysis focusing on long-term survival benefits may be reported in the near future. Therefore, a subsequent updated network meta-analysis will help to integrate the effects of these regimens on clinical outcomes. Despite the aforementioned limitations, our study has for the first time systematically compared the pCR efficacy and hematological SAEs of the currently available neoadjuvant chemotherapy regimens in TNBC. In conclusion, this study demonstrated that Za-, BP-, P-, and B-containing regimens are the top four treatment strategies, showing a high efficacy in achieving pCR. When considering both efficacy and SAEs, B-containing regimens had the highest acceptability to be the best treatment for achieving a


relatively high pCR with fewer SAEs. However, additional well-designed RCTs with larger sample sizes are required to strengthen the findings of this meta-analysis and further determine the survival benefits of these neoadjuvant therapies in TNBC. METHODS Search strategy This meta-analysis was performed in accordance with the guideline of Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [45]. Literatures, published before May 31, 2018, were identified in PubMed, Embase, Cochrane library, and Web of Science. The following key words were used: (breast OR mammary) AND (cancer OR cancers OR tumor OR neoplasm OR carcinoma) AND (neoadjuvant chemotherapy OR induction chemotherapy OR pre-operative chemotherapy) AND (TNBC OR triple-negative OR triple negative OR basal-like OR HER2 negative) AND (pathological complete response OR pCR), without any restrictions. The reference lists from relevant studies, reviews, and meta-analyses were manually screened for potentially eligible publications. Selection criteria The inclusion and exclusion criteria were prespecified. Eligible trials were studies prospectively comparing at least two arms of different neoadjuvant chemotherapy regimens in TNBC patients. Studies were excluded if they were: (1) non-human studies, letters, reviews, editorial comments or case reports; (2) single-arm or dosage-finding studies; (3) articles without raw data or with a sample size of less than 30; and (4) ongoing trials without reported results. If several publications from the same trial were identified, only the latest or complete publication was included. Two reviewers independently evaluated the risk of bias for eligible studies using the Cochrane Collaboration risk of bias tool [46]. Any discrepancies were resolved by discussing with all investigators. Data extraction Data were extracted by two independent authors. The following information was recorded for eligible studies: first author name, publication year, country, study design, trial phase, recruitment period, masking, number of centers, TNBC definition, clinical stage of patients, and trial name or registry number. Treatment regimens, sample size, primary outcome (pCR), and secondary outcome (grade 3–4 hematological adverse events), if available, were also recorded.

Treatment regimens and outcomes definition Due to the widespread use of combinations of taxanes, anthracyclines, cyclophosphamide, and/or fluorouracil, we defined the regimen as standard chemotherapeutic agents, which consisted of treatment with certain or all four drugs. The addition of any other drugs based on the standard chemotherapeutic agents was regarded as a new regimen. In the case of multi-arm studies comparing different dosages of one agent with another, we combined the results of the same agent into one arm. Administration sequence and frequency were not considered in the present study. The primary outcome was pCR, which was defined as the absence of invasive breast cancer in the breast and axillary lymph nodes (ypT0/is ypN0). If ypT0/is ypN0 was not reported, ypT0/is, defined as the absence of invasive breast cancer in the breast, was substituted. Secondary outcomes were treatment-related serious (grade 3–4) adverse events (SAEs) in hematology, including anemia, neutropenia, and thrombocytopenia. As for SAEs, we hypothesized that there was no significant correlation between molecular subtypes and hemato-logical adverse events. Therefore, if the SAEs of TNBC were not available, we included the number of SAEs from all molecular subtypes of breast cancer patients. If only the percentages of primary and secondary outcomes were reported, we calculated the number of events by multiplying percentages by the number of patients. Statistical methods The odds ratio (OR) was utilized for estimating pooling effect sizes. For the pairwise meta-analysis, hetero-geneity was calculated using Cochrane Q statistics and an I2 test. Statistical heterogeneity was defined as p<0.1 and/or I2>50%. A pairwise meta-analysis was conducted with a random-effects model or a fixed-effect model depending on the existence of statistical heterogeneity or not, respectively. All pairwise meta-analyses were performed by Review Manager software version 5.2. Results were reported as OR and 95% confidence intervals (CIs). All p-values were two-sided and a value less than 0.05 was considered statistically significant. Publication bias was detected by funnel plots, Begg’s and Egger’s tests using Stata software version 12.0. A Bayesian network-meta analysis was performed by Aggregate Data Drug Information System (ADDIS) software, version 1.16.8 (http://www. drugis.org) [47]. The consistency between direct and indirect evidence was verified by node splitting analyses. If no significant inconsistency was detected, a consistency model was used to analyze the relative effects of the interventions.


Otherwise, an inconsistency model was applied. Convergence was assessed using the Brooks-Gelman-Rubin method, which compared within-chain and between-chain variance to calculate the potential scale reduction factor (PSRF). We defined a PSRF of less than 1.05 as an indication of achieving convergence, and finished the simulation. Otherwise, the model should be extended until the PSRF is less than 1.05. The results of the network meta-analyses were presented as OR and 95% CI. Rank probability for each treatment was calculated to achieve the treatment ranking. Benefit-risk analyses were performed with synthesis evidence by stochastic multi-criteria acceptability analysis (SMAA) to jointly analyze the efficacy and SAEs for interventions. Rank acceptability and confidence factor (CF) were calculated with missing preference. AUTHOR CONTRIBUTIONS Conceptualization, G.S.R. and H.Z.L.; Methodology, Y.H.L. and D.J.Y.; Formal Analysis, Y.H.L. and P.C.; Investigation, Y.H.L., D.J.Y. and P.C.; Data Curation, H.Z.L. and X.D.Y.; Writing—Original Draft Preparation, Y.H.L., D.J.Y. and P.C.; Writing—Review & Editing, Y.H.L., D.J.Y., J.Z.S. and X.D.Y.; Supervision, G.S.R. and H.Z.L.; Project Administration, H.Z.L. All the authors read and approved the final manuscript. CONFLICTS OF INTEREST The authors declare no conflicts of interest. FUNDING This study was supported by National Natural Science Foundation of China (No. 81472475), Chongqing Science and Technology Commission (No. cstc2016jcyjA0313), and Scientific Research Foundation of Chongqing Medical University (No. 201408). REFERENCES 1. Foulkes WD, Smith IE, Reis-Filho JS. Triple-negative

breast cancer. N Engl J Med. 2010; 363:1938–48. https://doi.org/10.1056/NEJMra1001389

PMID:21067385

2. Joensuu H, Gligorov J. Adjuvant treatments for triple-negative breast cancers. Ann Oncol. 2012 (Suppl 6); 23:vi40–45.

https://doi.org/10.1093/annonc/mds194 PMID:23012301

3. Berry DA, Cirrincione C, Henderson IC, Citron ML, Budman DR, Goldstein LJ, Martino S, Perez EA, Muss HB, Norton L, Hudis C, Winer EP. Estrogen-receptor

status and outcomes of modern chemotherapy for patients with node-positive breast cancer. JAMA. 2006; 295:1658–67.

https://doi.org/10.1001/jama.295.14.1658 PMID:16609087

4. Carey LA, Dees EC, Sawyer L, Gatti L, Moore DT, Collichio F, Ollila DW, Sartor CI, Graham ML, Perou CM. The triple negative paradox: primary tumor chemosensitivity of breast cancer subtypes. Clin Cancer Res. 2007; 13:2329–34.

https://doi.org/10.1158/1078-0432.CCR-06-1109 PMID:17438091

5. Cortazar P, Zhang L, Untch M, Mehta K, Costantino JP, Wolmark N, Bonnefoi H, Cameron D, Gianni L, Valagussa P, Swain SM, Prowell T, Loibl S, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet. 2014; 384:164–72.

https://doi.org/10.1016/S0140-6736(13)62422-8 PMID:24529560

6. Liedtke C, Mazouni C, Hess KR, André F, Tordai A, Mejia JA, Symmans WF, Gonzalez-Angulo AM, Hennessy B, Green M, Cristofanilli M, Hortobagyi GN, Pusztai L. Response to neoadjuvant therapy and long-term survival in patients with triple-negative breast cancer. J Clin Oncol. 2008; 26:1275–81.

https://doi.org/10.1200/JCO.2007.14.4147 PMID:18250347

7. Ando M, Yamauchi H, Aogi K, Shimizu S, Iwata H, Masuda N, Yamamoto N, Inoue K, Ohono S, Kuroi K, Hamano T, Sukigara T, Fujiwara Y. Randomized phase II study of weekly paclitaxel with and without carboplatin followed by cyclophosphamide/epirubicin/5-fluorouracil as neoadjuvant chemotherapy for stage II/IIIA breast cancer without HER2 overexpression. Breast Cancer Res Treat. 2014; 145:401–09.

https://doi.org/10.1007/s10549-014-2947-1 PMID:24728578

8. Zhang P, Yin Y, Mo H, Zhang B, Wang X, Li Q, Yuan P, Wang J, Zheng S, Cai R, Ma F, Fan Y, Xu B. Better pathologic complete response and relapse-free survival after carboplatin plus paclitaxel compared with epirubicin plus paclitaxel as neoadjuvant chemotherapy for locally advanced triple-negative breast cancer: a randomized phase 2 trial. Oncotarget. 2016; 7:60647–56.

https://doi.org/10.18632/oncotarget.10607 PMID:27447966

9. Loibl S, O’Shaughnessy J, Untch M, Sikov WM, Rugo HS, McKee MD, Huober J, Golshan M, von Minckwitz G, Maag D, Sullivan D, Wolmark N, McIntyre K, et al. Addition of the PARP inhibitor veliparib plus carboplatin or carboplatin alone to standard

https://doi.org/10.1056/NEJMra1001389

https://www.ncbi.nlm.nih.gov/pubmed/21067385

https://doi.org/10.1093/annonc/mds194


https://doi.org/10.1001/jama.295.14.1658


https://doi.org/10.1158/1078-0432.CCR-06-1109


https://doi.org/10.1016/S0140-6736%2813%2962422-8


https://doi.org/10.1200/JCO.2007.14.4147


https://doi.org/10.1007/s10549-014-2947-1


https://doi.org/10.18632/oncotarget.10607



neoadjuvant chemotherapy in triple-negative breast cancer (BrighTNess): a randomised, phase 3 trial. Lancet Oncol. 2018; 19:497–509.


10. Earl HM, Hiller L, Dunn JA, Blenkinsop C, Grybowicz L, Vallier AL, Abraham J, Thomas J, Provenzano E, Hughes-Davies L, Gounaris I, McAdam K, Chan S, et al, and ARTemis Investigators. Efficacy of neoadjuvant bevacizumab added to docetaxel followed by fluorouracil, epirubicin, and cyclophosphamide, for women with HER2-negative early breast cancer (ARTemis): an open-label, randomised, phase 3 trial. Lancet Oncol. 2015; 16:656–66.


11. Rugo HS, Olopade OI, DeMichele A, Yau C, van ’t Veer LJ, Buxton MB, Hogarth M, Hylton NM, Paoloni M, Perlmutter J, Symmans WF, Yee D, Chien AJ, et al, and I-SPY 2 Investigators. Adaptive Randomization of Veliparib-Carboplatin Treatment in Breast Cancer. N Engl J Med. 2016; 375:23–34.

https://doi.org/10.1056/NEJMoa1513749 PMID:27406347

12. Llombart-Cussac A, Bermejo B, Villanueva C, Delaloge S, Morales S, Balmaña J, Amillano K, Bonnefoi H, Casas A, Manso L, Roché H, Gonzalez-Santiago S, Gavilá J, et al. SOLTI NeoPARP: a phase II randomized study of two schedules of iniparib plus paclitaxel versus paclitaxel alone as neoadjuvant therapy in patients with triple-negative breast cancer. Breast Cancer Res Treat. 2015; 154:351–57.


13. Bafeta A, Trinquart L, Seror R, Ravaud P. Reporting of results from network meta-analyses: methodological systematic review. BMJ. 2014; 348:g1741.

https://doi.org/10.1136/bmj.g1741 PMID:24618053

14. Enriquez D, Nieto NP, Fuentes HA, Guerra H, Mendoza RE, Gomez HL. Improving pathological response in locally advanced triple negative breast cancer: comparison between CbD and AC-T regimens. J Clin Oncol. 2017; 35:35.

https://doi.org/10.1200/JCO.2017.35.15_suppl.585

15. Aft R, Naughton M, Trinkaus K, Watson M, Ylagan L, Chavez-MacGregor M, Zhai J, Kuo S, Shannon W, Diemer K, Herrmann V, Dietz J, Ali A, et al. Effect of zoledronic acid on disseminated tumour cells in women with locally advanced breast cancer: an open label, randomised, phase 2 trial. Lancet Oncol. 2010; 11:421–28.


16. Huober J, von Minckwitz G, Denkert C, Tesch H, Weiss E, Zahm DM, Belau A, Khandan F, Hauschild M, Thomssen C, Högel B, Darb-Esfahani S, Mehta K, Loibl S. Effect of neoadjuvant anthracycline-taxane-based chemotherapy in different biological breast cancer phenotypes: overall results from the GeparTrio study. Breast Cancer Res Treat. 2010; 124:133–40.


17. Bernsdorf M, Ingvar C, Jörgensen L, Tuxen MK, Jakobsen EH, Saetersdal A, Kimper-Karl ML, Kroman N, Balslev E, Ejlertsen B. Effect of adding gefitinib to neoadjuvant chemotherapy in estrogen receptor negative early breast cancer in a randomized phase II trial. Breast Cancer Res Treat. 2011; 126:463–70.


18. Alba E, Chacon JI, Lluch A, Anton A, Estevez L, Cirauqui B, Carrasco E, Calvo L, Segui MA, Ribelles N, Alvarez R, Sanchez-Muñoz A, Sanchez R, et al. A randomized phase II trial of platinum salts in basal-like breast cancer patients in the neoadjuvant setting. Results from the GEICAM/2006-03, multicenter study. Breast Cancer Res Treat. 2012; 136:487–93.

https://doi.org/10.1007/s10549-012-2100-y PMID:23053638

19. Gerber B, Loibl S, Eidtmann H, Rezai M, Fasching PA, Tesch H, Eggemann H, Schrader I, Kittel K, Hanusch C, Kreienberg R, Solbach C, Jackisch C, et al, and German Breast Group Investigators. Neoadjuvant vacizumab and anthracycline-taxane-based chemotherapy in 678 triple-negative primary breast cancers; results from the geparquinto study (GBG 44). Ann Oncol. 2013; 24:2978–84.

https://doi.org/10.1093/annonc/mdt361 PMID:24136883

20. Earl HM, Vallier AL, Hiller L, Fenwick N, Young J, Iddawela M, Abraham J, Hughes-Davies L, Gounaris I, McAdam K, Houston S, Hickish T, Skene A, et al, and Neo-tAnGo Investigators. Effects of the addition of gemcitabine, and paclitaxel-first sequencing, in neoadjuvant sequential epirubicin, cyclophosphamide, and paclitaxel for women with high-risk early breast cancer (Neo-tAnGo): an open-label, 2×2 factorial randomised phase 3 trial. Lancet Oncol. 2014; 15:201–12.


21. Gonzalez-Angulo AM, Akcakanat A, Liu S, Green MC, Murray JL, Chen H, Palla SL, Koenig KB, Brewster AM, Valero V, Ibrahim NK, Moulder-Thompson S, Litton JK, et al. Open-label randomized clinical trial of standard

https://doi.org/10.1016/S1470-2045%2818%2930111-6


https://doi.org/10.1016/S1470-2045%2815%2970137-3


https://doi.org/10.1056/NEJMoa1513749


https://doi.org/10.1007/s10549-015-3616-8


https://doi.org/10.1136/bmj.g1741


https://doi.org/10.1200/JCO.2017.35.15_suppl.585

https://doi.org/10.1016/S1470-2045%2810%2970054-1


https://doi.org/10.1007/s10549-010-1103-9


https://doi.org/10.1007/s10549-011-1352-2


https://doi.org/10.1007/s10549-012-2100-y


https://doi.org/10.1093/annonc/mdt361


https://doi.org/10.1016/S1470-2045%2813%2970554-0



neoadjuvant chemotherapy with paclitaxel followed by FEC versus the combination of paclitaxel and everolimus followed by FEC in women with triple receptor-negative breast cancer. Ann Oncol. 2014; 25:1122–27.

https://doi.org/10.1093/annonc/mdu124 PMID:24669015

22. Steger GG, Greil R, Lang A, Rudas M, Fitzal F, Mlineritsch B, Hartmann BL, Bartsch R, Melbinger E, Hubalek M, Stoeger H, Dubsky P, Ressler S, et al, and Austrian Breast and Colorectal Study Group (ABCSG). Epirubicin and docetaxel with or without capecitabine as neoadjuvant treatment for early breast cancer: final results of a randomized phase III study (ABCSG-24). Ann Oncol. 2014; 25:366–71.

https://doi.org/10.1093/annonc/mdt508 PMID:24347519

23. von Minckwitz G, Schneeweiss A, Loibl S, Salat C, Denkert C, Rezai M, Blohmer JU, Jackisch C, Paepke S, Gerber B, Zahm DM, Kümmel S, Eidtmann H, et al. Neoadjuvant carboplatin in patients with triple-negative and HER2-positive early breast cancer (GeparSixto; GBG 66): a randomised phase 2 trial. Lancet Oncol. 2014; 15:747–56.


24. Hasegawa Y, Tanino H, Horiguchi J, Miura D, Ishikawa T, Hayashi M, Takao S, Kim SJ, Yamagami K, Miyashita M, Konishi M, Shigeoka Y, Suzuki M, et al, and JONIE Study Group. Randomized Controlled Trial of Zoledronic Acid plus Chemotherapy versus Chemotherapy Alone as Neoadjuvant Treatment of HER2-Negative Primary Breast Cancer (JONIE Study). PLoS One. 2015; 10:e0143643.

https://doi.org/10.1371/journal.pone.0143643 PMID:26633806

25. Martinez MC, Arce-Salinas C, Alvarado-Miranda A, Lara-Medina F, Flores-Diaz D, Matus JA, Bargallo-Rocha E, Shaw-Dulin R, Maldonado H, Mendoza-Galindo L, Perez-Sanchez V. Randomized phase II trial to evaluate the safety and efficacy of neoadjuvant cisplatin in combination with taxanesanthracyclines vs taxanesanthracyclines alone in locally advanced triple negative breast cancer. J Clin Oncol. 2015; 33e12024–e12024. https://doi.org/10.1200/jco.2015.33.15_suppl.e12024

26. Sikov WM, Berry DA, Perou CM, Singh B, Cirrincione CT, Tolaney SM, Kuzma CS, Pluard TJ, Somlo G, Port ER, Golshan M, Bellon JR, Collyar D, et al. Impact of the addition of carboplatin and/or bevacizumab to neoadjuvant once-per-week paclitaxel followed by dose-dense doxorubicin and cyclophosphamide on pathologic complete response rates in stage II to III

triple-negative breast cancer: CALGB 40603 (Alliance). J Clin Oncol. 2015; 33:13–21.

https://doi.org/10.1200/JCO.2014.57.0572 PMID:25092775

27. Nahleh ZA, Barlow WE, Hayes DF, Schott AF, Gralow JR, Sikov WM, Perez EA, Chennuru S, Mirshahidi HR, Corso SW, Lew DL, Pusztai L, Livingston RB, Hortobagyi GN. SWOG S0800 (NCI CDR0000636131): addition of bevacizumab to neoadjuvant nab-paclitaxel with dose-dense doxorubicin and cyclophosphamide improves pathologic complete response (pCR) rates in inflammatory or locally advanced breast cancer. Breast Cancer Res Treat. 2016; 158:485–95.


28. Gluz O, Nitz U, Liedtke C, Christgen M, Grischke EM, Forstbauer H, Braun M, Warm M, Hackmann J, Uleer C, Aktas B, Schumacher C, Bangemann N, et al. Comparison of Neoadjuvant Nab-Paclitaxel+Carboplatin vs Nab-Paclitaxel+Gemcitabine in Triple-Negative Breast Cancer: Randomized WSG-ADAPT-TN Trial Results. J Natl Cancer Inst. 2018; 110:628–637.

https://doi.org/10.1093/jnci/djx258 PMID:29228315

29. Jovanović B, Mayer IA, Mayer EL, Abramson VG, Bardia A, Sanders ME, Kuba MG, Estrada MV, Beeler JS, Shaver TM, Johnson KC, Sanchez V, Rosenbluth JM, et al. A Randomized Phase II Neoadjuvant Study of Cisplatin, Paclitaxel With or Without Everolimus in Patients with Stage II/III Triple-Negative Breast Cancer (TNBC): Responses and Long-term Outcome Correlated with Increased Frequency of DNA Damage Response Gene Mutations, TNBC Subtype, AR Status, and Ki67. Clin Cancer Res. 2017; 23:4035–45.


30. Wu X, Tang P, Li S, Wang S, Liang Y, Zhong L, Ren L, Zhang T, Zhang Y. A randomized and open-label phase II trial reports the efficacy of neoadjuvant lobaplatin in breast cancer. Nat Commun. 2018; 9:832.


31. Poggio F, Bruzzone M, Ceppi M, Pondé NF, La Valle G, Del Mastro L, de Azambuja E, Lambertini M. Platinum-based neoadjuvant chemotherapy in triple-negative breast cancer: a systematic review and meta-analysis. Ann Oncol. 2018; 29:1497–508.

https://doi.org/10.1093/annonc/mdy127 PMID:29873695

32. Petrelli F, Coinu A, Borgonovo K, Cabiddu M, Ghilardi M, Lonati V, Barni S. The value of platinum agents as neoadjuvant chemotherapy in triple-negative breast cancers: a systematic review and meta-analysis. Breast Cancer Res Treat. 2014; 144:223–32.

https://doi.org/10.1093/annonc/mdu124


https://doi.org/10.1093/annonc/mdt508


https://doi.org/10.1016/S1470-2045%2814%2970160-3


https://doi.org/10.1371/journal.pone.0143643


https://doi.org/10.1200/jco.2015.33.15_suppl.e12024

https://doi.org/10.1200/JCO.2014.57.0572


https://doi.org/10.1007/s10549-016-3889-6


https://doi.org/10.1093/jnci/djx258


https://doi.org/10.1158/1078-0432.CCR-16-3055


https://doi.org/10.1038/s41467-018-03210-2


https://doi.org/10.1093/annonc/mdy127



https://doi.org/10.1007/s10549-014-2876-z PMID:24557340

33. Sharma P, López-Tarruella S, García-Saenz JA, Ward C, Connor CS, Gómez HL, Prat A, Moreno F, Jerez-Gilarranz Y, Barnadas A, Picornell AC, Del Monte-Millán M, Gonzalez-Rivera M, et al. Efficacy of Neoadjuvant Carboplatin plus Docetaxel in Triple-Negative Breast Cancer: Combined Analysis of Two Cohorts. Clin Cancer Res. 2017; 23:649–57.


34. Sonnenblick A, de Azambuja E, Azim HA Jr, Piccart M. An update on PARP inhibitors—moving to the adjuvant setting. Nat Rev Clin Oncol. 2015; 12:27–41.

https://doi.org/10.1038/nrclinonc.2014.163 PMID:25286972

35. Miller K, Wang M, Gralow J, Dickler M, Cobleigh M, Perez EA, Shenkier T, Cella D, Davidson NE. Paclitaxel plus bevacizumab versus paclitaxel alone for metastatic breast cancer. N Engl J Med. 2007; 357:2666–76.


36. Chen H, Lu W, Zhang Y, Zhu X, Zhou J, Chen Y. A Bayesian network meta-analysis of the efficacy of targeted therapies and chemotherapy for treatment of triple-negative breast cancer. Cancer Med. 2019; 8:383–99. https://doi.org/10.1002/cam4.1892

PMID:30525293

37. Ge L, Tang Y, Zhang QN, Tian JH, Wang XH, Pieper D, Pan B, Li L, Ling J, Bing ZT, Yang KH. A network meta-analysis on the efficacy of targeted agents in combination with chemotherapy for treatment of advanced/metastatic triple-negative breast cancer. Oncotarget. 2017; 8:59539–51.

https://doi.org/10.18632/oncotarget.19102 PMID:28938657

38. Kimmel DB. Mechanism of action, pharmacokinetic and pharmacodynamic profile, and clinical applications of nitrogen-containing bisphosphonates. J Dent Res. 2007; 86:1022–33.

https://doi.org/10.1177/154405910708601102 PMID:17959891

39. Jagdev SP, Coleman RE, Shipman CM, Rostami-H A, Croucher PI. The bisphosphonate, zoledronic acid, induces apoptosis of breast cancer cells: evidence for synergy with paclitaxel. Br J Cancer. 2001; 84:1126–34.

https://doi.org/10.1054/bjoc.2001.1727 PMID:11308265

40. Kunzmann V, Bauer E, Feurle J, Weissinger F, Tony HP, Wilhelm M. Stimulation of gammadelta T cells by aminobisphosphonates and induction of antiplasma cell activity in multiple myeloma. Blood. 2000; 96:384–92.

PMID:10887096

41. Santini D, Vincenzi B, Dicuonzo G, Avvisati G, Massacesi C, Battistoni F, Gavasci M, Rocci L, Tirindelli MC, Altomare V, Tocchini M, Bonsignori M, Tonini G. Zoledronic acid induces significant and long-lasting modifications of circulating angiogenic factors in cancer patients. Clin Cancer Res. 2003; 9:2893–97.

PMID:12912933

42. Ottewell PD, Mönkkönen H, Jones M, Lefley DV, Coleman RE, Holen I. Antitumor effects of doxorubicin followed by zoledronic acid in a mouse model of breast cancer. J Natl Cancer Inst. 2008; 100:1167–78.

https://doi.org/10.1093/jnci/djn240 PMID:18695136

43. Gnant M, Mlineritsch B, Schippinger W, Luschin-Ebengreuth G, Pöstlberger S, Menzel C, Jakesz R, Seifert M, Hubalek M, Bjelic-Radisic V, Samonigg H, Tausch C, Eidtmann H, et al, and ABCSG-12 Trial Investigators. Endocrine therapy plus zoledronic acid in premenopausal breast cancer. N Engl J Med. 2009; 360:679–91.


44. Gnant M, Mlineritsch B, Stoeger H, Luschin-Ebengreuth G, Heck D, Menzel C, Jakesz R, Seifert M, Hubalek M, Pristauz G, Bauernhofer T, Eidtmann H, Eiermann W, et al, and Austrian Breast and Colorectal Cancer Study Group, Vienna, Austria. Adjuvant endocrine therapy plus zoledronic acid in premenopausal women with early-stage breast cancer: 62-month follow-up from the ABCSG-12 randomised trial. Lancet Oncol. 2011; 12:631–41.

https://doi.org/10.1016/S1470-2045(11)70122-X PMID:21641868

45. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, Clarke M, Devereaux PJ, Kleijnen J, Moher D. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. PLoS Med. 2009; 6:e1000100.

https://doi.org/10.1371/journal.pmed.1000100 PMID:19621070

46. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, Savovic J, Schulz KF, Weeks L, Sterne JA, and Cochrane Bias Methods Group, and Cochrane Statistical Methods Group. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ. 2011; 343:d5928. https://doi.org/10.1136/bmj.d5928 PMID:22008217

47. Van Valkenhoef G, Tervonen T, Zwinkels T, De Brock B, Hillege H. ADDIS: A decision support system for evidence-based medicine. Decis Support Syst. 2013; 55:459–75.

https://doi.org/10.1016/j.dss.2012.10.005

https://doi.org/10.1007/s10549-014-2876-z


https://doi.org/10.1158/1078-0432.CCR-16-0162


https://doi.org/10.1038/nrclinonc.2014.163




https://doi.org/10.1002/cam4.1892


https://doi.org/10.18632/oncotarget.19102


https://doi.org/10.1177/154405910708601102


https://doi.org/10.1054/bjoc.2001.1727




https://doi.org/10.1093/jnci/djn240




https://doi.org/10.1016/S1470-2045%2811%2970122-X


https://doi.org/10.1371/journal.pmed.1000100


https://doi.org/10.1136/bmj.d5928


https://doi.org/10.1016/j.dss.2012.10.005


SUPPLEMENTARY MATERIALS Supplementary Figures

Supplementary Figure 1. Funnel plots of the publication bias tests for direct comparisons of pCR. (A) Standard chemotherapeutic agents vs. P-containing regimens. (B) Standard chemotherapeutic agents vs. B-containing regimens.

Supplementary Figure 2. Subgroup Bayesian network meta-analysis for pathologic complete response (pCR). (A) The league table of comparisons. Data are presented as odds radio (OR) and 95% confidence intervals (CI). An OR>1 favors the column-defining treatment, and OR <1 favors the row-defining treatment. (B) Heatmap of the rank probability of the regimens for pCR. Rank 1 represents the best reatment and rank 10 represents the worst. Rank probabilities sum to one, both within a rank over treatments and within a treatment over ranks.


Supplementary Figure 3. Network diagram of eligible comparisons included in the network meta-analysis for serious adverse events. (A) Network diagram of eligible comparisons for anemia. (B) Network diagram of eligible comparisons for neutropenia. (C) Network diagram of eligible comparisons for thrombocytopenia. The node size is proportional to the total number of patients in the regimen. The width of each line is proportional to the number of studies comparing the two regimens linked by the line. B, bevacizumab; P, platinum salts; BP, bevacizumab plus platinum salts; Pi, Poly (ADP-ribose) polymerases inhibitors; PPi, platinum salts plus Poly (ADP-ribose) polymerases inhibitors; Ca, capecitabine; Ge, gemcitabine; Za, zoledronic acid; E, everolimus; PE, platinum salts plus everolimus; G, gefitinib.


Supplementary Figure 4. Pairwise meta-analyses of serious adverse events. (A) Forest plot for anemia. Standard chemotherapeutic agents vs. P-containing regimens. (B) Forest plot for neutropenia. Standard chemotherapeutic agents vs. P-containing regimens. (C) Forest plot for neutropenia. Standard chemotherapeutic agents vs. B-containing regimens. (D) Forest plot for thrombocytopenia. Standard chemotherapeutic agents vs. P-containing regimens.


Supplementary Figure 5. Subgroup pairwise meta-analyses of serious adverse events in TNBC patients. (A) Forest plot for anemia. (B) Forest plot for neutropenia. (C) Forest plot for thrombocytopenia. The three comparisons are standard chemotherapeutic agents vs. P-containing regimens.


Supplementary Figure 6. Funnel plots of the publication bias tests for direct comparisons of serious adverse events. (A) Funnel plot for anemia. (B) Funnel plot for neutropenia. (C) Funnel plot for thrombocytopenia. The three comparisons are standard chemotherapeutic agents vs. P-containing regimens.


Supplementary Figure 7. Heatmaps of the rank probability of the regimens for serious adverse events. (A) Rank probability of anemia. (B) Rank probability of neutropenia. (C) Rank probability of thrombocytopenia. Rank 1 represents the worst treatment and rank N represents the best. Rank probabilities sum to one, both within a rank over treatments and within a treatment over ranks.


Supplementary Figure 8. Central weights and confidence factor (CF) for each regimen. (A) Analysis based on synthesizing pCR and anemia. (B) Analysis based on synthesizing pCR and neutropenia. (C) Analysis based on synthesizing pCR and thrombocytopenia. (D) Analysis based on synthesizing pCR and the three serious adverse events. The sum of the central weight for each intervention equals to one. A high central weight indicates that the intervention being considered is the best for that outcome. The confidence factor represents the probability of an intervention obtaining the first rank.


Supplementary Tables Please browse Full Text version to see the data of:

Supplementary Table 1. Treatment characteristics of the eligible studies.

Supplementary Table 2. Risk of bias assessments for included trials.

Study

Selection bias-

Random sequence

generation

Selection bias-

Allocation concealment

Performance bias-Blinding

of participants

and personnel

Detection bias-

Blinding of outcome

assessment

Attrition bias-

Incomplete outcome

data

Reporting bias-

Selective reporting

Other bias

Aft 2010 + + + + + + ? Houber 2010 ? ? + + + + + Bernsdorf 2011 ? ? + + + + ? Alba 2012 + + + + + + + Gerber 2013 ? ? + + + + + Ando 2014 ? ? + + + + + Earl 2014 + + + + + + + Gonzalez-Angulo 2014 + + + + + + +

Steger 2014 + ? + + + + + von Minckwitz 2014 + + + + + + +

Earl 2015 + + + + + + + Hasegawa 2015 + + + + + + ? Llombart-Cussac 2015 ? ? + + + + +

Martinez 2015 ? ? + + + ? + Sikov 2015 ? ? + + + + + Nahleh 2016 ? ? + + + + + Zhang 2016 ? ? + + + + + Rugo 2016 + + + + + + + Enriquez 2017 - - + + + + + Gluz 2017 ? + + + + + + Jovanović 2017 + + + + + + + Loibl 2018 + + + + + + +

Wu 2018 + ? + + + + +

+=low risk of bias; -=high risk of bias; ?=unclear risk of bias. Other bias is defined as the definition of pCR (i.e., whether it is defined clearly).


Supplementary Table 3. Grade 3-4 hematological adverse events.

Study Arm design NACT regimens BC patients No. of

patients Anemia Neutropenia Thrombocytopenia

Aft 2010 Standard

chemotherapeutic agents vs. Za-

containing regimens

TA NA NA NA NA NA

TAZa NA NA NA NA NA

Houber 2010

Standard chemotherapeutic

agents vs. Ca-containing regimens

TAC NA NA NA NA NA

TACCa NA NA NA NA NA

Bernsdorf 2011


agents vs. G-containing regimens

AC ER- NA NA NA NA

ACG ER- NA NA NA NA

Alba 2012 Standard

chemotherapeutic agents vs. P-containing

regimens

TAC TNBC 46 0 10 0

TACP TNBC 47 5 8 3

Gerber 2013


agents vs. B-containing regimens

TAC Her2- 963 16/963 747/939 15/963

TACB Her2- 951 17/950 763/936 20/951

Steger 2014


agents vs. Ca-containing regimens

TA NA NA NA NA NA

TACa NA NA NA NA NA

Ando 2014


agents vs. P-containing regimens

TACF Her2- 91 1 35 0

TACFP Her2- 88 17 58 1

Earl 2014 Standard

chemotherapeutic agents vs. Ge-

containing regimens

TAC All 404 3 135 2

TACGe All 408 0 146 0

Gonzalez-Angulo 2014


agents vs. E-containing regimens

TACF TNBC 27 1 11 2

TACFE TNBC 23 4 12 1

von Minckwitz 2014

B-containing regimens vs. BP-containing

regimens

TAB TNBC/Her2+ 293 1 79 1 TABP TNBC/Her2+ 295 45 192 42

Earl 2015 Standard

chemotherapeutic agents vs. B-containing

regimens

TACF All 391 NA 146 NA

TACFB All 384 NA 168 NA

Hasegawa 2015


agents vs. Za-containing regimens

TACF Her2- 95 NA 34 NA

TACFZa Her2- 91 NA 39 NA

Llombart-Cussac 2015


agents vs. Pi-containing regimens

T TNBC 46 NA 2 NA

TPi TNBC 94 NA 5 NA

Martinez 2015



TACF NA NA NA NA NA

TAP NA NA NA NA NA

Sikov 2015


agents vs. B-containing regimens vs. P-

containing regimens vs.

TAC TNBC 107 0 24 4 TACB TNBC 105 2 28 3 TACP TNBC 111 4 62 22

TACBP TNBC 110 6 74 29


BP-containing regimens

Nahleh 2016


agents vs. B-containing regimens

TAC NA NA NA NA NA

TACB NA NA NA NA NA

Zhang 2016



TA TNBC 44 NA 28 0

TP TNBC 47 NA 34 17

Rugo 2016


agents vs. PPi-containing regimens

TAC Her2- 44 0 1 0

TACPPi Her2- 72 20 51 15

Enriquez 2017



TAC NA NA NA NA NA

TP NA NA NA NA NA

Gluz 2017 Ge-containing regimens vs. P-

containing regimens

TGe TNBC 180 NA 29 2 TP TNBC 151 NA 24 0

Jovanović 2017

P-containing regimens vs. PE-containing

regimens

TP TNBC 49 1 5 0 TPE TNBC 96 6 25 0

Loibl 2018


agents vs. P-containing regimens vs. PPi-

containing regimens

TAC TNBC 157 0 4 0 TACP TNBC 158 27 84 10

TACPPi TNBC 313 77 179 33

Wu 2018 Standard

chemotherapeutic agents vs. P-containing

regimens

TA TNBC 60 6 14 1

TAP TNBC 61 32 24 21

A, anthracyclines; B, bevacizumab; C, cyclophosphamide; Ca, capecitabine; E, everolimus; F, florouracil; G, gefitinib; Ge, gemcitabine; P, platinum salts; Pi,Poly(ADP-ribose) polymerases (PARPs); T, taxanes; Za, zoledronic acid; NA, not available.

Documents

Efficacy and safety of neoadjuvant chemotherapy …...after treatment have a high risk of recurrence [5]. Although there is a significant association between pCR and survival outcomes,